scripts/validate_items.py

from __future__ import annotations

from pathlib import Path

from local_hardware_items import (
    ItemValidationError,
    catalog_from_items,
    load_items,
    load_memory_price_sources,
)

ROOT = Path(__file__).resolve().parents[1]
ITEMS_DIR = ROOT / "data" / "items"
MEMORY_PRICE_SOURCES_PATH = ROOT / "data" / "memory_price_sources.csv"
BANDS = [
    ("under_3000", "<$3,000", 0, 3000),
    ("3000_5000", "$3,000–$5,000", 3000, 5000),
    ("5000_10000", "$5,000–$10,000", 5000, 10000),
    ("10000_25000", "$10,000–$25,000", 10000, 25000),
    ("25000_75000", "$25,000–$75,000", 25000, 75000),
    ("75000_plus", "$75,000+", 75000, None),
]


def main() -> int:
    try:
        items = load_items(ITEMS_DIR, bands=BANDS)
        catalog = catalog_from_items(items)
        memory_price_rows = load_memory_price_sources(
            MEMORY_PRICE_SOURCES_PATH,
            items,
        )
    except ItemValidationError as exc:
        print("Item validation failed.")
        print(exc)
        return 1

    item_count = len(items)
    bandwidth_items = {
        row["item_id"]
        for row in catalog
        if row.get("bandwidth_gbps") != ""
    }
    missing_bandwidth = sorted(
        path.stem for path in ITEMS_DIR.glob("*.json")
        if path.stem not in bandwidth_items
    )
    print(f"Validated {item_count} item files and {len(catalog)} catalog rows.")
    print(f"Memory price rows: {len(memory_price_rows)}.")
    print(f"Bandwidth values: {len(bandwidth_items)}/{item_count} item files.")
    if missing_bandwidth:
        print("Missing bandwidth values:")
        for item_id in missing_bandwidth:
            print(f"- {item_id}")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())